Word-Level Correction of Speech Input

1. A computer-implemented method comprising: providing a plurality of transcribed words from a word lattice for display on a computing device; in response to receiving an indication that a particular transcribed word has been selected, providing for display on the computing device a particular alternate phrase from the word lattice, wherein the particular alternate phrase includes alternate words that correspond to (i) the particular transcribed word and (ii) at least one transcribed word preceding the particular transcribed word; and in response to receiving an indication that the particular alternate phrase has been selected, replacing, with the particular transcribed phrase, (i) the particular transcribed word and (ii) the at least one transcribed word preceding the particular transcribed word.

2. The method of claim 1 , wherein providing the plurality of transcribed words from the word lattice for display on the computing device comprises: receiving speech audio data from a microphone in the computing device; transmitting the speech audio data to a transcription system hosted by the computing device; and receiving the word lattice transcribed from the speech audio data by the transcription system.

3. The method of claim 1 , wherein providing the plurality of transcribed words from the word lattice for display on the computing device comprises: receiving speech audio data from a microphone in the computing device; transmitting the speech audio data to a transcription system that is in communication with the computing device over a network; and receiving the word lattice transcribed from the speech audio data by the transcription system.

4. The method of claim 1 , wherein the word lattice comprises nodes corresponding to the transcribed words and words included in the particular alternate phrase, edges between the nodes that identify possible paths through the word lattice, and each path has an associated probability of being correct.

5. The method of claim 4 , further comprising identifying the particular alternate phrase for the particular transcribed word and at least one transcribed word preceding the particular transcribed word from at least one alternate path between a beginning node and an ending node of the presented transcribed words in the word lattice.

6. The method of claim 1 , wherein receiving the indication that the particular transcribed word has been selected and receiving the indication that the particular alternate phrase has been selected comprise receiving the indication that the particular transcribed word has been selected and receiving the indication that the particular alternate phrase has been selected through a touchscreen interface of the computing device by a user of the computing device.

7. The method of claim 1 , wherein the word lattice is one of: a hierarchical word lattice, or a finite state transducer word lattice.

8. A computer program product, encoded on a non-transitory computer-readable medium, operable to cause one or more processors to perform operations for correcting words in transcribed text, the operations comprising: providing a plurality of transcribed words from a word lattice for display on a computing device; in response to receiving an indication that a particular transcribed word has been selected, providing for display on the computing device a particular alternate phrase from the word lattice, wherein the particular alternate phrase includes alternate words that correspond to (i) the particular transcribed word and (ii) at least one transcribed word preceding the particular transcribed word; and in response to receiving an indication that the particular alternate phrase has been selected, replacing, with the particular transcribed phrase, (i) the particular transcribed word and (ii) the at least one transcribed word preceding the particular transcribed word.

9. The computer program product of claim 8 , wherein providing the plurality of transcribed words from the word lattice for display on the computing device comprises: receiving speech audio data from a microphone in the computing device; transmitting the speech audio data to a transcription system hosted by the computing device; and receiving the word lattice transcribed from the speech audio data by the transcription system.

10. The computer program product of claim 8 , wherein providing the plurality of transcribed words from the word lattice for display on the computing device comprises: receiving speech audio data from a microphone in the computing device; transmitting the speech audio data to a transcription system that is in communication with the computing device over a network; and receiving the word lattice transcribed from the speech audio data by the transcription system.

11. The computer program product of claim 8 , wherein the word lattice comprises nodes corresponding to the transcribed words and words included in the particular alternate phrase, edges between the nodes that identify possible paths through the word lattice, and each path has an associated probability of being correct.

12. The computer program product of claim 11 , further comprising identifying the particular alternate phrase for the particular transcribed word and at least one transcribed word preceding the particular transcribed word from at least one alternate path between a beginning node and an ending node of the presented transcribed words in the word lattice.

13. The computer program product of claim 8 , wherein receiving the indication that the particular transcribed word has been selected and receiving the indication that the particular alternate phrase has been selected comprise receiving the indication that the particular transcribed word has been selected and receiving the indication that the particular alternate phrase has been selected through a touchscreen interface of the computing device by a user of the computing device.

14. The computer program product of claim 8 , wherein the word lattice is one of: a hierarchical word lattice, or a finite state transducer word lattice.

15. A system for correcting words in transcribed text, the system comprising: one or more computers; and a computer-readable medium coupled to the one or more computers having instructions stored thereon which, when executed by the one or more computers, cause the one or more computers to perform operations comprising: providing a plurality of transcribed words from a word lattice for display on a computing device; in response to receiving an indication that a particular transcribed word has been selected, providing for display on the computing device a particular alternate phrase from the word lattice, wherein the particular alternate phrase includes alternate words that correspond to (i) the particular transcribed word and (ii) at least one transcribed word preceding the particular transcribed word; and in response to receiving an indication that the particular alternate phrase has been selected, replacing, with the particular transcribed phrase, (i) the particular transcribed word and (ii) the at least one transcribed word preceding the particular transcribed word.

16. The system of claim 15 , wherein the operations of providing the plurality of transcribed words from the word lattice for display on the computing device comprise: receiving speech audio data from a microphone in the computing device; transmitting the speech audio data to a transcription system hosted by the computing device; and receiving the word lattice transcribed from the speech audio data by the transcription system.

17. The system of claim 15 , wherein the operations of providing the plurality of transcribed words from the word lattice for display on the computing device comprise: receiving speech audio data from a microphone in the computing device; transmitting the speech audio data to a transcription system that is in communication with the computing device over a network; and receiving the word lattice transcribed from the speech audio data by the transcription system.

18. The system of claim 15 , wherein the word lattice comprises nodes corresponding to the transcribed words and words included in the particular alternate phrase, edges between the nodes that identify possible paths through the word lattice, and each path has an associated probability of being correct.

19. The system of claim 18 , further comprising identifying the particular alternate phrase for the particular transcribed word and at least one transcribed word preceding the particular transcribed word from at least one alternate path between a beginning node and an ending node of the presented transcribed words in the word lattice.

20. The system of claim 15 , wherein receiving the indication that the particular transcribed word has been selected and receiving the indication that the particular alternate phrase has been selected comprise receiving the indication that the particular transcribed word has been selected and receiving the indication that the particular alternate phrase has been selected through a touchscreen interface of the computing device by a user of the computing device.